chromatin interactions predict co-expression in the mouse ... de... · ahmed mahfouz1,2*, sepideh...

1
Summary Conclusion We find that the chromatin inter action prole of a gene-pair is a good predictor of their spatial co-expression. The accuracy of the prediction substantially improves when chromatin interac- tions are described by scale-aware topological measures of the multi-resolution chromatin inter action network. For co-expres- sion prediction, it is necessary to take into account different levels of chromatin inter actions from direct inter action between genes (i.e. small-scale) to chromatin compartment interac- tions (i.e. large-scale). Co-expression Prediction Topological Measures The 3D conf ormation of the genome in the cell nucleus influ- ences important biological processes such as gene expression. Predicting gene co-expression from frequent long-r ange chro- matin inter actions is a challenging task. We address this by char - acterizing the topology of the cortical chromatin inter action net - work using scale-aware topological measures 1 . We demon- strate that based on these char acterizations it is possible to ac- curately predict spatial co-expression between genes in the mouse cortex. of Chromatin Interaction Network of Chromosome 16 Gene A Gene B Gene C Co-expression Network Chromatin Interaction Network Gene D 3D Structure of Genome Gene A Gene B Gene C Gene D Coronal View of Cortical Layers Gene A Gene B Gene C Mid Layers Deep Layer Expression Level Low High Gene D Gene C Gene A Gene B Gene D Prediction Shortest Path Jaccard Index Scale-aware Standard Hi-C Resolution Measure 200 kb Standard Standard Multi-resolution Scale-aware Scale-aware 200 kb Multi-resolution 0.65 0.7 0.75 0.8 0.85 Chr1 ChrX Topological Signature 0 1 B3galt5 - Carhsp1 Thpo Kcnj6 Igsf11 Hmox2 Gpr156 Fstl1 Ece2 Dyrk1a D16H22S680E Clcn2 Chrd Carhsp1 B3galt5 Sidt1 Robo1 Tagln3 Zbtb20 Nde1 Abat Fam55c Masp1 Cd47 0 1 Scale-aware Jaccard Index (at Medium-Scale) Abat - Masp1 Reference [1] M. Hulsman, et al., Graph-topological scale-spaces reveal functional information in physical protein interaction networks, Bioinformatics (2014) Chromatin Interaction Network: Intrachromosomal Hi-C contacts from mouse cortical cells 2 Co-expression Network: Spatial co-expression between genes from Allen Mouse Brain Atlas (ABA) 3 Scale-aware topological measures [2] Y . Shen, et al., A map of the cis-regulatory sequences in the mouse genome. Nature (2012) [3] E. Lein, et al., Genome-wide atlas of gene expression in the adult mouse brain. Nature (2007) Chromatin interactions predict co-expression in the mouse cortex Ahmed Mahfouz 1,2* , Sepideh Babaei 1* , Marc Hulsman 1,3 , Boudewijn PF Lelieveldt 2 , Jeroen de Ridder 1 , Marcel Reinders 1 Co-expression Chromatin Interaction 1 Delft Bioinformatics Lab, Delft University of Technology, 2 Dept. of Radiology, Leiden University Medical Center, 3 Department of Clinical Genetics, VU University Medical Center *Shared first authorship, [email protected]

Upload: others

Post on 08-Aug-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Chromatin interactions predict co-expression in the mouse ... de... · Ahmed Mahfouz1,2*, Sepideh Babaei1*, Marc Hulsman1,3, Boudewijn PF Lelieveldt2, Jeroen de Ridder1, Marcel Reinders1

Summary

ConclusionWe find that the chromatin interaction profile of a gene-pair is a good predictor of their spatial co-expression. The accuracy of the prediction substantially improves when chromatin interac-tions are described by scale-aware topological measures of the multi-resolution chromatin interaction network. For co-expres-sion prediction, it is necessary to take into account different levels of chromatin interactions from direct interaction between genes (i.e. small-scale) to chromatin compartment interac-tions (i.e. large-scale).

Co-expression Prediction

Topological Measures

The 3D conformation of the genome in the cell nucleus influ-ences important biological processes such as gene expression. Predicting gene co-expression from frequent long-range chro-matin interactions is a challenging task. We address this by char-acterizing the topology of the cortical chromatin interaction net-work using scale-aware topological measures1. We demon-strate that based on these characterizations it is possible to ac-curately predict spatial co-expression between genes in the mouse cortex.

of Chromatin Interaction Network of Chromosome 16

Gene AGene B

Gene C

Co-expression Network

Chromatin Interaction Network

Gene D

3D Structure of Genome

Gene AGene B

Gene C

Gene D

Coronal View of Cortical Layers

Gene A Gene B

Gene CMid Layers

Deep Layer

Expression LevelLow High

Gene D

Gene C

Gen

e A

Gen

e B

Gen

e D

Pred

icti

on

Shortest Path

Jaccard Index

Scale-awareStandard

Hi-C Resolution

Measure

200 kb

Standard Standard

Multi-resolution

Scale-aware Scale-aware

200 kb Multi-resolution

0.65

0.7

0.75

0.8

0.85 Chr1 ChrX

Topological Signature

0 1

B3galt5 - Carhsp1 Thpo

Kcnj6

Igsf11Hmox2

Gpr156

Fstl1

Ece2

Dyrk1a

D16H22S680E

Clcn2

Chrd

Carhsp1B3galt5

Sidt1

Robo1

Tagln3

Zbtb20

Nde1 Abat

Fam55c

Masp1

Cd47

0 1

Scale-aware Jaccard Index (at Medium-Scale)

Abat - Masp1

Reference[1] M. Hulsman, et al., Graph-topological scale-spaces reveal functional information in physical protein interaction networks, Bioinformatics (2014)

Chromatin Interaction Network: Intrachromosomal Hi-C contacts from mouse cortical cells2

Co-expression Network:Spatial co-expression between genes from Allen Mouse Brain Atlas (ABA)3

Scale-aware topological measures

[2] Y. Shen, et al., A map of the cis-regulatory sequences in the mouse genome. Nature (2012)[3] E. Lein, et al., Genome-wide atlas of gene expression in the adult mouse brain. Nature (2007)

Chromatin interactions predict co-expression in the mouse cortex

Ahmed Mahfouz1,2*, Sepideh Babaei1*, Marc Hulsman1,3, Boudewijn PF Lelieveldt2, Jeroen de Ridder1, Marcel Reinders1

Co-expressionChromatin Interaction

1Delft Bioinformatics Lab, Delft University of Technology, 2Dept. of Radiology, Leiden University Medical Center, 3Department of Clinical Genetics, VU University Medical Center*Shared first authorship, [email protected]